TFS-18528; No of Pages 9 Technological Forecasting & Social Change xxx (2016) xxx–xxx

Contents lists available at ScienceDirect

Technological Forecasting & Social Change

Environmental catching-up, eco-innovation, and technological leadership in China's pilot ecological civilization zones

Yanni Yu a,WenjieWub, Tao Zhang c,⁎,YanchuLiud a Institute of Resource, Environment and Sustainable Development, College of Economics, Jinan University, Guangzhou, Guangdong 510632, China b School of Energy, Geoscience, Infrastructure and Society, Heriot-Watt University, Edinburgh EH14 4AS, UK c Department of Marketing, Birmingham Business School, University of Birmingham, Edgbaston, Birmingham B15 2TT, UK d Lingnan College, Sun Yat-Sen University, Guangzhou 510275, China article info abstract

Article history: In this study we propose a global metafrontier Luenberger productivity indicator (GMLPI) to investigate the effect Received 24 December 2015 of the establishment of the Poyang Lake Eco-economic Zone (PLEEZ), one of China's typical ecological civilization Received in revised form 5 May 2016 zones, on regional environmental total-factor productivity growth. We combine the global environmental tech- Accepted 14 May 2016 nology, metafrontier approach, and the non-radial Luenberger productivity indicator and incorporate the region- Available online xxxx al heterogeneities into the environmental productivity growth analysis. This GMLPI includes the efficiency change, technological change, and metafrontier technology gap change indices. An empirical study of the Keywords: Environmental total-factor productivity PLEEZ has been conducted using the county level data covering a period from 2009 to 2013. Empirical results Metafrontier non-radial Luenberger show that the environmental productivity growth has increased by 8.71% on average, with growth primarily productivity indicator driven by technological change. These results suggest that the establishment of the PLEEZ is effective in encour- Poyang lake Eco-economic Zone aging eco-innovation; however the PLEEZ lacks an eco-leadership effect. Significant heterogeneities in environ- mental productivity growth and its patterns among three major functional zones in the PLEEZ remain. © 2016 Elsevier Inc. All rights reserved.

1. Introduction widely used for measuring productivity change growth. Taking the environmental factors into account, Chung et al. (1997) proposed The Poyang Lake Eco-economic Zone Planning was officially ap- the Malmquist–Luenberger index for measuring environmentally proved by the State Council of China on December 12th, 2009, indicating sensitive productivity growth. Empirical studies applying the that the construction of the Poyang Lake Eco-economic Zone (PLEEZ) Malmquist–Luenberger index include Weber and Domazlicky was a national-level strategy. The PLEEZ is based on the Poyang lake (2001), Färe et al. (2001), Yörük and Zaim (2005), Kumar (2006) urban circle and is intended to become an ecological economic demon- and Zhang and Choi (2013a,b), for measuring environmental perfor- stration zone and a low-carbon economy development priority zone in mance change. China.1 The increasing interest in and strategic importance of the PLEEZ The Luenberger productivity indicator is an alternative for measur- have heightened the need for investigating its effects on regional envi- ing productivity growth. Compared with the Luenberger indicator, the ronmental performance. Malmquist index appears to overestimate productivity changes Therefore this study investigates the effect of the PLEEZ's establish- (Boussemart et al., 2003). Recent work also suggests that the ment on regional environmental productivity growth after 2009. A sig- Luenberger index is more robust than the Malmquist index (Fujii nificant growth of the environmental productivity of the PLEEZ after et al., 2014). Thus, this study employs the Luenberger indicator as its 2009 indicates that the PLEEZ policy is effective. main methodology. However, the basic Luenberger indicator cannot Two kinds of indices are widely used for measuring productivity deal with the non-zero slack variable problem because it adopts the ra- growth, i.e. the Malmquist index and the Luenberger indicator. The dial method for measuring performance (Zhou et al., 2012; Zhang et al., Malmquist index, empirically presented by Färe et al. (1994),is 2014). To handle this shortcoming, Fujii et al. (2014) proposed the non- radial Luenberger indicator with undesirable outputs. To incorporate cross-group heterogeneity into this indicator, Zhang and Wei (2015) in- troduced a metafrontier non-radial Luenberger carbon emission perfor- ⁎ Corresponding author. E-mail addresses: [email protected] (Y. Yu), [email protected] (T. Zhang). mance index; unlike Zhang and Wei (2015), who focused on single- 1 For details about the PLEEZ, please see Appendix A. factor carbon emission performance change, we propose a metafrontier

http://dx.doi.org/10.1016/j.techfore.2016.05.010 0040-1625/© 2016 Elsevier Inc. All rights reserved.

Please cite this article as: Yu, Y., et al., Environmental catching-up, eco-innovation, and technological leadership in China's pilot ecological civilization zones, Technol. Forecast. Soc. Change (2016), http://dx.doi.org/10.1016/j.techfore.2016.05.010 2 Y. Yu et al. / Technological Forecasting & Social Change xxx (2016) xxx–xxx non-radial Luenberger productivity indicator to measure the total- to scale may then be expressed as follows: factor environmentally sensitive productivity growth. Compared with the methodologies adopted in existing energy and environmental stud- XN ; ; : ≤ ; ; :::; ; ies (e.g. Wang et al., 2013, 2015; Zhang and Wei, 2015),2 the key inno- T ¼fðÞx y b znxmn xm m ¼ 1 M n¼1 vativeness of the proposed approach is that it can handle both non-zero XN slack variables and cross-group heterogeneity simultaneously. We then z y ≥y ; s ¼ 1; :::; S; n sn s ð2Þ use the proposed approach to measure environmental productivity n¼1 XN growth and its patterns in the PLEEZ at a county level. z b ¼ b ; j ¼ 1; :::; J; This study has two major contributions to the current literature. n jn j n¼1 Practically, although a number of studies have focused on the environ- zn ≥0; n ¼ 1; ⋯; Ng: mental productivity growth for China's provinces (Zhang et al., 2011; Choi et al., 2015) or China's industries (Chen and Golley, 2014; Li and where M, S and J denote the number of inputs, desirable outputs and un- Lin, 2015), no studies have explored the PLEEZ, a national-level strategy. desirable outputs, respectively. This study thus addresses this research gap by evaluating the PLEEZ, After defining the environmental technology, the non-radial direc- thus enabling us to get insight into the effectiveness of this national- tional distance function (DDF) is followed. The formal definition of the level strategy. Methodologically, we develop an integrated methodolo- non-radial DDF was first introduced by Zhou et al. (2012). Compared gy called the Global Metafrontier non-radial Luenberger Productivity In- with other energy and environmental modeling techniques, a unique dicator (GMLPI) to measure environmental productivity growth. This advantage of DDF is its capability of expanding desirable outputs and methodology combines a non-radial directional distance function, lessening inputs or undesirable outputs simultaneously. A review of Luenberger productivity indicator, and the global metafrontier ap- DDF application in energy and environmental studies can be found in proach. It can thus incorporate slack variables, undesirable outputs, Zhang and Choi (2014). Following Zhou et al. (2012), the non-radial and group heterogeneities when measuring environmental productivi- DDF with undesirable output is defined as: ty growth. The remainder of this paper is organized as follows. Section 2 intro- ! D ðÞ¼x; y; b; g sup wT β : ðÞðÞþx; y; b g diagðÞβ ∈T : ð3Þ duces the methodology used. Section 3 conducts the empirical study. Section 4 concludes this study. x y b T where w =(wm,ws,wj ) denotes a normalized weight vector corre- sponding to the number of inputs and outputs, g =(−gx,gy,−gb)isan β βx βy βb T ≥ 2. Methodology explicit directional vector, and =( m, s, j ) 0denotesthescaling ! factors indicating the inefficiencies. Thus the value of D ðx; y; b; gÞ 2.1. Non-radial directional distance function under the environmental technology can be calculated by solving the following DEA-type model: We assume that there are j =1,…, N observations, which are differ- ∈ℜM ! x y b ent counties in the PLEEZ, and each region uses input vector x + to D ðÞx; y; b; g ¼ max wx β þ wyβ þ wbβ ∈ℜS m m s s j j produce good economic output vector y + as well as undesirable XN pollutant vector b ∈ℜJ . Thus, the environmental production technolo- : : ≤ −βx ; ; :::; ; + s t znxmn xm mgxm m ¼ 1 M gy can be expressed as follows: n¼1 XN ≥ βy ; ;::: ; znysn ys þ s gys s ¼ 1 S ð4Þ ; ; : ; ; n¼1 T ¼ fgðÞx y b x can produce ðÞy b ð1Þ XN −βb ; ; ::: ; znbjn ¼ b j j gbj j ¼ 1 J n¼1 ≥ ; ; ; ⋯; where T is the environmental production technology. We assume that it zn 0 n ¼ 1 2 N βx ; βy; βb ≥ : satisfies the standard axioms of production theory (Färe and Grosskopf, m s j 0 2005). That is, finite amounts of inputs can produce only finite outputs. Inputs and desirable outputs are often assumed to be freely (strong) dis- The directional vector g can be set up in different ways, based on posable. For modeling joint-production technologies (Färe et al., 1989) ! given policy goals. If D ðx; y; b; gÞ¼0, then the specificunittobeevalu- with undesirable outputs, weak disposability and null-jointness as- ated is located on the frontier of the best practices in the direction of g. sumptions should be imposed on the environmental production tech- nology T, as follows: 2.2. Global metafrontier Luenberger productivity indicator (GMLPI)

(i) If (x,y,b) ∈ T and 0 ≤ θ ≤ 1, then (x,θy,θb) ∈ T; Following Zhang and Choi (2013a,b) and Zhang and Wei (2015), (ii) If (x,y,b) ∈ T and b =0,theny =0. three environmental technologies: contemporaneous, intertemporal, and global environmental technologies are needed to define the GMLPI and its decompositions. The weak-disposability assumption implies that pollutant abate- The contemporaneous environmental technology for group Rh at ment activities are costly in terms of proportional reductions in product fi c t t t t t t time t is de ned as TRh ={(x ,y ,b ):(x ) can produce (y ,b )}, where output. Meanwhile, the null-jointness assumption indicates that the t =1,…,T. The intertemporal environmental production technology of pollutants are not avoidable in the PLEEZ unless economic activities fi I 1 ∪ 2 ∪ ∪ T group Rh is de ned as TRh = TRh TRh ... TRh .Theintertemporalen- are stopped. vironmental technology can be interpreted as the single technology set Following the literature, a piecewise non-parametric linear frontier that encompasses all contemporaneous environmental technologies is adopted to construct the environmental production technology. The from whole period only for the specific group Rh. The global environ- environmental technology T for N observations with constant returns fi G I ∪ I ∪ ∪ I mental production technology is de ned as T = TR1 TR2 ... TRH , which is constructed from all observations over the whole period for all groups. This implies that the global environmental technology en- 2 For a comprehensive review please see Zhang and Choi (2014). compasses all intertemporal environmental production technologies,

Please cite this article as: Yu, Y., et al., Environmental catching-up, eco-innovation, and technological leadership in China's pilot ecological civilization zones, Technol. Forecast. Soc. Change (2016), http://dx.doi.org/10.1016/j.techfore.2016.05.010 Y. Yu et al. / Technological Forecasting & Social Change xxx (2016) xxx–xxx 3

Table 1 models: Descriptive statistics. !d s; s; s; βd βd βd Variable Unit Mean St. dev. Min Max D x y b g X¼ maxwx x þ wy y þ wb b d 8 : : s s ≤ −β GDP Good output 10 Yuan 159.1 133.5 20.5 814.9 s t znxn xn0 x gx Waste water Bad output 104 ton 2774.5 2993.9 280.8 25,921.7 Xcon s s ≥ βd SO2 Bad output ton 8994.3 13,270.7 41.2 58,524.1 znyn yn0 þ ygy ð5Þ Soot Bad output ton 3570.5 8091.3 20.3 94,012.8 Xcon 8 s s −βd Capital Input 10 Yuan 570.6 413.7 54.2 2516.6 znbn ¼ bn0 bgb Labor Input 104 27.7 19.4 3.8 94.0 con 4 s ≥ ; βd ≥ Energy Input 10 ton 75.3 124.4 0.1 671.7 zn 0 0

! Here the superscript d on D dðxs; ys; bs; gÞ represents the different and it is assumed that all observations can access the global technology type of NDDF, which can be contemporaneous, intertemporal, or global. through their eco-innovation. The symbol con under ∑ represents the conditions for constructing the The contemporaneous non-radial DDF is constructed based on the three environmental production technologies. The contemporaneous c fi NDDF should follow the conditions d ≡ C and con ≡ {n∈R }; the contemporaneous environmental production technology (TRh )ofspeci c h !I intertemporal NDDF, d ≡ I and con ≡ {n∈Rh,s∈[1,2, …,T]}; and for the group Rh. Similarly, the intertemporal non-radial DDF is defined asD ð:Þ¼ global NDDF, d ≡ G and con ≡ {n∈[R1 ∪ R2 ∪ ... ∪ RH], s∈[1,2, …,T]}. supfwT βI : ððx; y; bÞþg diagðβIÞÞ∈TI g based on the intertemporal en- Rh Following the idea of the Luenberger indicator which is based on the vironmental production technology (T I)ofgroupR . Finally the global Rh h DDF, the GMLPI is defined as follows ! DDF: D Gð:Þ¼ supfwT βG : ððx; y; bÞþg diagðβGÞÞ∈TGg is based on the ! ! global environmental production technology (TG). GMLPI xS; yS; bS ¼ D G xtþ1; ytþ1; btþ1 − D G xt ; yt ; bt ð6Þ To calculate and decompose the GMLPI, we need to solve six differ- ! ! ! ent functions: D C ðxs; ys; bsÞ , D Iðxs; ys; bsÞ,and D Gðxs; ys; bsÞ, S = t, The GMLPI in Eq. (6) measures the environmental productivity t + 1. The NDDFs can be computed by using the following DEA-type growth using an additive measure based on the differences between

Fig. 1. Illustration of Major functional Zones in the PLEEZ.

Please cite this article as: Yu, Y., et al., Environmental catching-up, eco-innovation, and technological leadership in China's pilot ecological civilization zones, Technol. Forecast. Soc. Change (2016), http://dx.doi.org/10.1016/j.techfore.2016.05.010 4 Y. Yu et al. / Technological Forecasting & Social Change xxx (2016) xxx–xxx

Table 2 Basic statistics for three major functional zones.

Group GDP (108 Yuan) Energy (104 ton) Capital (108 Yuan) Labor (104) SO2 (ton)

Mean Growth Mean Growth Mean Growth Mean Growth Mean Growth

KDZ 214.3167 14.78% 111.7367 3.90% 706.2642 16.72% 28.58847 4.77% 13,120.09 3.72% MAPZ 84.94987 16.60% 26.8112 1.04% 380.2558 9.88% 33.07815 0.68% 3241.925 6.40% KEPZ 53.5446 15.68% 4.4205 4.61% 327.0234 7.37% 12.77227 1.74% 1520.433 3.43% PLEEZ 117.6037 15.69% 47.65614 3.18% 471.1811 11.33% 24.81296 2.40% 5960.818 4.52% two DDFs. From Eq. (6), the GMLPI measures the observation of move- during the period from 2009 to 2013. First, we describe how the data ment (toward or away from) of the global environmental production pos- for the empirical study was collected. sibilities frontier from period t to t + 1. GMLPI N 0 indicates that the For the output variables, because our study focuses on regional envi- environmental productivity growth has been improved, and then the ronmental economies, we choose regional real GDP to represent the observation is moving toward the global environmental frontier. If only desirable output. For the input variables, we choose labor, capital GMLPI = 0, the productivity does not change. If GMLPI b0, the environ- and energy consumption data. We use the number of employed labor mental productivity drops and the observation is moving away from the force as the labor data. Because the capital stock data is unavailable, global frontier. we use the growth rate approach to estimate the capital stock for each The GMLPI can be decomposed into a group technical efficiency county. The growth rate approach has been proved to be an effective change (EC) index, a group technological change (TC) index, and a one in Young (2003). metafroniter technology gap change (MGC) index of environmental Based on this approach, the capital stock can be calculated as: productivity. The decomposition process is as follows: Δ Kn;tþ1 K ; ¼ ð8Þ S S S !G tþ1 tþ1 tþ1 !G t t t n t δ GMLPI x ; y ; b ¼ D x ; y ; b − D x ; y ; b ðÞn þ gn h hi ! ! D tþ1 xtþ1; ytþ1; btþ1 − D t xt; yt; bt ¼ where Kn,t denotes the capital stock of region n in time t, ΔKn,t +1is the hi investment in capital of region n in time t +1,andδn and gn indicate the !I tþ1; tþ1; tþ1 −!tþ1 tþ1; tþ1; tþ1 þ D x y b D x y b depreciation rate and GDP growth rate of region n, respectively. The hi ! ! data of δ for Jiangxi Province comes from Wu (2009). − D I xt ; yt ; bt − D t xt; yt; bt n 7 hi ð Þ ! ! þ D G xtþ1; ytþ1; btþ1 − D I xtþ1; ytþ1; btþ1 Table 3 Changes in the GMLPI in the 38 PLEEZ regions, 2009–2013. hi ! ! − D G xt; yt; bt − D I xt ; yt ; bt Regions Groups 09–10 10–11 11–12 12–13 Mean Changjiang District KDZ −0.0321 0.0844 0.0313 −0.1287 −0.0113 tþ1− t tþ1− t tþ1− t ¼ TE TE þ TC TC þ MGC MGC Donghu District KDZ 0.1289 0.4756 0.5794 0.5665 0.4376 ¼ EC þ TC þ MGC Fengcheng City KDZ 0.1651 −0.1665 0.0369 0.0027 0.0096 Gaoan City KDZ 0.1779 −0.1561 0.0241 0.0029 0.0122 The efficiency change (EC) index in Eq. (7) measures the “catch-up” Gongqincheng City KDZ 0.2365 0.2665 0.0407 0.3773 0.2302 Guixi City KDZ 0.052 0.0335 0.1033 0.0712 0.065 effect in terms of technical efficiency changes in environmental produc- Hukou County KDZ −0.0567 0.0402 −0.0011 0.0326 0.0037 tivity for a specific observation in a group during two periods (t, t +1). Jiujiang County KDZ 0.2161 −0.222 0.0134 −0.005 0.0006 EC captures how close an observation is moving toward the contempo- Leping City KDZ 0.0944 0.0447 0.0313 0.0339 0.0511 raneous environmental production possibilities frontier. Here EC N (or Linchuan District KDZ 0.1216 −0.0497 0.0347 0.0376 0.036 − − b) 0 means an efficiency gain (or loss). Lushan District KDZ 0.0074 0.0303 0.0664 0.0236 0.0013 Nanchang County KDZ 0.4824 0.9346 0.0673 0.2994 0.4459 The technological change (TC) index measures changes in the best- Pengze County KDZ 0.0719 −0.2907 0.0128 −0.0354 −0.0604 practice gap for the environmental technology between the contempo- Qingshanhu District KDZ −0.1203 0.0304 −0.0675 −0.0195 −0.0442 raneous environmental technology and the intertemporal environmen- Qingyunpu District KDZ −0.0876 0.3561 −0.023 0.0624 0.077 tal technology during two periods. TC N (or b) 0 means that the Ruichang City KDZ 0.0263 0.0177 0.0198 −0.0262 0.0094 Xihu District KDZ 0.1284 0.6217 0.4899 0.5858 0.4565 contemporaneous technology frontier shifts toward (or away from) Xinjian County KDZ 0.1035 0.2447 0.0468 0.1279 0.1307 the intertemporal technology frontier. Because TC measures frontier Xunyang District KDZ 0.305 0.248 0.0654 0.1558 0.1935 shifts in a contemporaneous technology, it can be considered to be a Yushui District KDZ 0.1301 −0.0368 −0.0251 0.0727 0.0352 measure of eco-innovation effect. Yuehu District KDZ 0.1114 0.7385 0.1531 0.1672 0.2926 − − − Metafrontier gap change (MGC) is a measure of changes in the Zhangshu City KDZ 0.1418 0.2544 0.0692 0.0223 0.0164 Zhushan District KDZ 0.151 0.1819 0.3417 0.2329 0.2269 metafrontier technology gap (MTG) for eco-frontier shift between the Dean County MAPZ 0.2592 0.0717 0.1484 0.1521 0.1578 intertemporal environmental production technology frontier and the Dongshan County MAPZ 0.1841 0.0532 0.0984 0.1316 0.1168 global frontier during two periods. TGC N (or b) 0 indicates a decrease Duchang County MAPZ 0.0496 −0.1909 0.0344 0.0039 −0.0257 (or increase) in the meta-technology gap between the intertemporal Jinxian County MAPZ 0.1528 0.2462 0.0989 0.0446 0.1356 Poyang County MAPZ 0.1015 −0.1046 0.0249 −0.0228 −0.0002 technology for a specific group and the global environmental technolo- Wannian County MAPZ −0.1184 0.2891 0.2998 0.1307 0.1503 gy. Therefore, MGC represents the eco-technological leadership change Xingan County MAPZ 0.1648 −0.1334 0.0764 0.0074 0.0288 for a given group. Yongxiu County MAPZ 0.0672 −0.1189 0.1854 −0.0282 0.0264 Yugan County MAPZ 0.3548 −0.2645 0.0992 −0.0337 0.0389 − − 3. Empirical analysis Yujiang County MAPZ 0.3646 0.0349 0.0191 0.0147 0.0914 Anyi County KEFZ 0.0799 0.0372 0.0128 0.0369 0.0417 Fuliang County KEFZ 0.0385 0.0201 −0.0707 0.0102 −0.0005 3.1. Data Wanli District KEFZ 0.0244 0.1595 −0.1003 0.0015 0.0212 Wuning County KEFZ 0.2181 −0.2689 0.017 −0.0075 −0.0103 The GMLPI explained in Section 2 is used to measure environmental Xingzi County KEFZ −0.3823 0.1517 0.1315 −0.0774 −0.0441 Mean PLEEZ 0.1083 0.083 0.0793 0.0776 0.0871 productivity growth and its patterns in the 38 counties in the PLEEZ

Please cite this article as: Yu, Y., et al., Environmental catching-up, eco-innovation, and technological leadership in China's pilot ecological civilization zones, Technol. Forecast. Soc. Change (2016), http://dx.doi.org/10.1016/j.techfore.2016.05.010 Y. Yu et al. / Technological Forecasting & Social Change xxx (2016) xxx–xxx 5

Table 4 principle of developing modern agriculture and ensuring the food secu- Group efficiency change component of the GMLPI, 2009–2013. rity. The KEFZs maintain the ecosystem for its essential ecological func- Regions Group 09–10 10–11 11–12 12–13 Mean tions. As shown in Table 3, in the PLEEZ, the KDZs include 23 counties, the MAPZs contain 10 counties, and the KZFZs include 5 counties. Fig. Changjiang District KDZ −0.0923 −0.2270 0.0121 −0.2074 −0.1286 Donghu District KDZ 0.0000 0.0000 0.0000 0.0000 0.0000 1 illustrates the geographical location of each county with its major Fengcheng City KDZ −0.0311 −0.4497 0.0374 −0.0658 −0.1273 functional zone classification information. Gaoan City KDZ 0.0880 −0.2899 0.0130 −0.0805 −0.0674 In order to illustrate the heterogeneity in the PLEEZ, we compare the Gongqincheng City KDZ 0.0000 0.0000 0.0000 0.0000 0.0000 inputs and outputs across the three major functional zones (Table 2). Guixi City KDZ −0.0242 −0.0263 0.0193 −0.0292 −0.0151 Hukou County KDZ −0.1293 −0.0384 −0.0028 0.0233 −0.0368 Because soot and wastewater are in a situation similar to that of SO2, Jiujiang County KDZ 0.0000 −0.2827 −0.0789 −0.0798 −0.1103 we only list the information for SO2. The KDZs have more capital and Leping City KDZ −0.0106 −0.0591 0.0068 −0.0308 −0.0234 consume greater amounts of energy to produce GDP. The other two Linchuan District KDZ 0.0530 −0.1901 0.0180 −0.0059 −0.0312 types of zones produce less GDP and use less energy and labor. The var- Lushan District KDZ −0.1146 −0.1168 −0.0803 −0.0012 −0.0782 iables thus illustrate the widening gap between the KDZs and the other Nanchang County KDZ 0.3302 0.0000 0.0000 0.0000 0.0825 Pengze County KDZ −0.1266 −0.3513 −0.0239 −0.0606 −0.1406 two types of zones. The KEFZs consume less energy and contribute less Qingshanhu District KDZ −0.1667 −0.0022 −0.0675 −0.0212 −0.0644 SO2 emissions than the MAPZs because they are focused on ensuring Qingyunpu District KDZ −0.1339 0.0891 −0.0554 0.0007 −0.0249 ecological security. Ruichang City KDZ −0.0799 −0.0959 −0.0371 −0.0797 −0.0731 Xihu District KDZ 0.0000 0.0000 0.0000 0.0000 0.0000 Xinjian County KDZ −0.0551 0.1064 0.0089 −0.0121 0.0120 3.3. Empirical results Xunyang District KDZ 0.0000 0.0000 0.0000 0.0000 0.0000 Yushui District KDZ 0.0000 −0.1211 −0.0995 0.0230 −0.0494 3.3.1. Environmental productivity growth Yuehu District KDZ 0.0674 0.0006 0.0000 0.0000 0.0170 We have calculated the GMLPI for each of the 38 regions in the PLEEZ Zhangshu City KDZ 0.0000 −0.2699 0.0123 −0.0816 −0.0848 to assess the environmental productivity change incorporating regional Zhushan District KDZ 0.0000 0.0000 0.0000 0.0000 0.0000 Dean County MAPZ 0.0041 0.0000 0.0000 0.0000 0.0010 heterogeneity. Table 3 shows the empirical results of the GMLPI index Dongshan County MAPZ 0.0282 0.0704 0.1479 0.1634 0.1025 for the 2009–2013 period. The GMLPI increased approximately 8.71% Duchang County MAPZ −0.1189 −0.0578 −0.1260 0.0994 −0.0508 from 2009 to 2013 on average, which indicates that the PLEEZ area Jinxian County MAPZ 0.0000 0.0000 0.0000 0.0000 0.0000 had a notable increase in environmental productivity after its establish- Poyang County MAPZ 0.0000 0.0000 0.0000 0.0000 0.0000 fi Wannian County MAPZ −0.0906 0.0906 0.0000 0.0000 0.0000 ment. Signi cant downward trends in the GMLPI appeared in 11 regions Xingan County MAPZ −0.0087 −0.1305 −0.0219 −0.0451 −0.0516 during the research period. Among those, Pengze county shows the Yongxiu County MAPZ −0.0167 −0.0454 0.0414 −0.0353 −0.0140 largest decrease in GMLPI (−0.06). Other regions demonstrate varying Yugan County MAPZ 0.0372 −0.2363 0.1023 0.0272 −0.0174 Yujiang County MAPZ 0.0000 0.0000 −0.1130 −0.1417 −0.0637 Table 5 Anyi County KEFZ −0.0851 −0.0057 0.0034 0.0390 −0.0121 Group technological change component of the GMLPI, 2009–2013. Fuliang County KEFZ 0.0000 0.0000 −0.0330 0.0040 −0.0073 Wanli District KEFZ 0.0000 0.0000 0.0000 0.0000 0.0000 Regions Group 09–10 10–11 11–12 12–13 Mean Wuning County KEFZ 0.0000 −0.2230 0.0555 −0.1022 −0.0674 Xingzi County KEFZ −0.1080 −0.0171 0.1251 −0.1594 −0.0398 Changjiang District KDZ 0.0602 0.3114 0.0192 0.0812 0.1180 Mean −0.0206 −0.0758 −0.0036 −0.0226 −0.0306 Donghu District KDZ 0.1289 0.4753 0.5794 0.5665 0.4375 Fengcheng City KDZ 0.2034 0.2592 0.0008 0.0574 0.1302 Gaoan City KDZ 0.0968 0.1264 0.0099 0.0770 0.0775 The integrated energy consumption is selected as the energy input. Gongqincheng City KDZ 0.2159 0.2636 0.0407 0.3773 0.2244 Guixi City KDZ 0.0720 0.0571 0.0705 0.1004 0.0750 All types of energy (coal, oil, and gas) are converted into the equivalent Hukou County KDZ 0.0745 0.0752 0.0017 0.0092 0.0402 of tons of standard coal. For undesirable outputs, SO2 emissions, waste- Jiujiang County KDZ 0.1633 0.1414 0.0461 0.0681 0.1047 water, and soot emissions are used, since these types of data are avail- Leping City KDZ 0.0388 0.0883 0.0131 0.0573 0.0494 able for the PLEEZ. We then collect the input and output data of the 38 Linchuan District KDZ 0.0667 0.1375 0.0069 0.0379 0.0623 counties for empirical analysis. All the data used are from the Jiangxi Sta- Lushan District KDZ 0.1123 0.1426 0.0139 0.0247 0.0733 Nanchang County KDZ 0.1412 0.9345 0.0671 0.2994 0.3606 tistical Year Book and Statistics of Poyang Lake Eco-Economic Zone. Table 1 Pengze County KDZ 0.1991 0.0989 0.0267 0.0217 0.0866 reports the descriptive statistics of these data. Qingshanhu District KDZ 0.0463 0.0326 0.0000 0.0017 0.0202 Qingyunpu District KDZ 0.0463 0.2670 0.0324 0.0624 0.1020 3.2. Regional heterogeneity in PLEEZ Ruichang City KDZ 0.0168 0.0941 0.0439 0.0462 0.0503 Xihu District KDZ 0.1284 0.6217 0.4899 0.5858 0.4565 Xinjian County KDZ 0.0497 0.1390 0.0303 0.1446 0.0909 The rapid economic development of China has impelled authorities Xunyang District KDZ 0.3050 0.2513 0.0733 0.1558 0.1963 to explore policies that prevent unreasonable land use, severe ecological Yushui District KDZ 0.1199 0.0864 0.0744 0.0497 0.0826 damage, and uncoordinated development of urban and rural areas. Yuehu District KDZ 0.0407 0.6737 0.1486 0.1672 0.2575 Under this background, the State Council of China has formulated the Zhangshu City KDZ 0.1165 0.1035 0.0320 0.0473 0.0748 Zhushan District KDZ 0.2179 0.3225 0.4438 0.2436 0.3069 Nationwide Major Function Oriented Zone policy. Dean County MAPZ 0.2742 0.2241 0.1785 0.2459 0.2307 In its development planning, the Jiangxi provincial government has Dongshan County MAPZ 0.1233 0.0399 0.1961 0.1474 0.1267 divided Jiangxi province into three major types of functional zones: Duchang County MAPZ 0.1664 0.0173 0.1275 0.0357 0.0867 key development zones (KDZ), limited development zones (LDZ), and Jinxian County MAPZ 0.1660 0.4084 0.4343 0.1614 0.2925 Poyang County MAPZ 0.1656 0.1451 0.0631 0.1070 0.1202 development prohibited zones (DPZ). Wannian County MAPZ 0.0219 0.1841 0.3351 0.1684 0.1774 fi To use the metafrontier approach, the rst step should be to charac- Xingan County MAPZ 0.2140 0.0685 0.1282 0.0866 0.1243 terize the groups and determine their members. In this paper, the Major Yongxiu County MAPZ 0.1343 0.1047 0.1714 0.1423 0.1382 Functional Zone classification is selected as the criteria for categorizing Yugan County MAPZ 0.3095 0.0274 0.0497 0.0119 0.0996 the groups based on their different development targets, i.e. economic Yujiang County MAPZ 0.4023 0.4524 0.1623 0.1015 0.2796 Anyi County KEFZ 0.1249 0.0357 0.0000 0.0376 0.0496 development and urbanization levels Accordingly, the counties were di- Fuliang County KEFZ 0.0599 0.0463 0.0027 0.0244 0.0333 vided into three types: key development zone (KDZ), major agriculture Wanli District KEFZ 0.2610 1.1047 0.1176 0.0670 0.3876 production zone (MAPZ) and key ecological function zone (KEFZ). The Wuning County KEFZ 0.2025 0.1164 0.0064 0.0346 0.0900 KDZs are urbanized areas with a high level of industrialization and Xingzi County KEFZ 0.0739 0.2481 0.0297 0.0525 0.1010 Mean PLEEZ 0.1411 0.2349 0.1123 0.1239 0.1530 high capital intensity. The MAPZs offer primary products with the

Please cite this article as: Yu, Y., et al., Environmental catching-up, eco-innovation, and technological leadership in China's pilot ecological civilization zones, Technol. Forecast. Soc. Change (2016), http://dx.doi.org/10.1016/j.techfore.2016.05.010 6 Y. Yu et al. / Technological Forecasting & Social Change xxx (2016) xxx–xxx degrees of growth. Xihu District, Nanchang County, and Donghu District show relatively high GMLPI growth rates. Average GMLPI grew yearly, especially during the 2009–2011 period (0.108). Interestingly, these re- sults are different from a recent study investigating the effect of the PLEEZ on agricultural labor productivity (Wu and Wang, 2015), where the authors found that the establishment of the PLEEZ had negative im- pact on the agricultural labor productivity. This difference can reflect a major function of the PLEEZ—developing economy in an eco-/ environmental friendly way—which should positively influence the en- vironmental total-factor productivity; however, developing economy in an eco-/environmental friendly way does not mean an increase in the agricultural labor productivity.

3.3.2. Group efficiency change Fig. 2. GMLPI trend and decompositions. The GMLPI is decomposed into three indices to investigate its growth patterns: group efficiency change (GEC), group technological change (GTC), and metafrontier gap change (MGC). highest GEC and the lowest one is Yujiang County (−6.37%). In the As shown in Table 4, the average group efficiency change (GEC) of KEFZ, each region shows no improvement in GEC, with the lowest the PLZZE is −0.0306 using the GMLPI estimate, suggesting that the en- beingWuningCounty(−6.74%). vironmental efficiency decreased approximately 3.06% per year during 2009–2013. At the regional level, 24 regions show a dip in GEC, while 3.3.3. Group technological change GEC grows only among 5 regions. In the other 9 regions, GEC is largely Table 5 shows that the average GTC for all observations in the sample unchanged. Among all regions, the highest growth rate, 10%, is in period is 0.1530, indicating that the regions experienced a significant Dongshan County. The growth rate of GEC in Pengze County is the low- improvement in technology, 15.3% on average. Only Qingshanhu Dis- est (−14.1%), indicating the county lags in environmental efficiency trict and Anyi County remained unchanged in technology during performance. 2011–2012. A number of regions exhibited a sharp increase in techno- For the different functional zones, Nanchang County in a KDZ has the logical change during 2010–2011, especially Wanli District (1.105), highest GEC, with an average of 8.25%, whereas the lowest GEC is Yuehu District (0.674), and Xihu District (0.622). marked by Pengze County. In the MAPZ, Dongshan County shows the We can conclude that the establishment of the PLEEZ stimulated eco-innovation in environmental-friendly production processes for Table 6 most of the counties in the PLEEZ area. Metafrontier gap change component of the GMLPI, 2009–2013.

Regions Cluster 09–10 10–11 11–12 12–13 Mean 3.3.4. Metafrontier gap change Metafrontier gap change (MGC) measures the disparity between the Changjiang District KDZ 0.0000 0.0000 0.0000 −0.0025 −0.0006 Donghu District KDZ 0.0000 0.0004 0.0000 0.0000 0.0001 environmental frontier of the PLEEZ and the environmental frontiers of Fengcheng City KDZ −0.0072 0.0240 −0.0013 0.0110 0.0066 the three groups. Thus, it can measure the technological heterogeneity Gaoan City KDZ −0.0069 0.0074 0.0012 0.0064 0.0020 across the three major functional zones. Table 6 shows that the average Gongqincheng City KDZ 0.0206 0.0028 0.0000 0.0000 0.0059 MGC is −0.035, implying that the technology gap between the group- Guixi City KDZ 0.0042 0.0027 0.0135 0.0000 0.0051 frontier and metafrontier has widened by 3.5%. This may show that Hukou Prefecture KDZ −0.0020 0.0034 0.0000 0.0000 0.0004 Jiujiang Prefecture KDZ 0.0528 −0.0807 0.0462 0.0067 0.0062 the PLEEZ region significantly lacks technological leadership effects dur- Leping City KDZ 0.0662 0.0155 0.0114 0.0074 0.0251 ing the sample period. From the functional zone perspective, there were Linchuan District KDZ 0.0019 0.0029 0.0098 0.0055 0.0050 no technological leadership effects in the MAPZs, while only one region Lushan District KDZ 0.0097 0.0045 0.0000 0.0001 0.0036 (Anyi) in the KEFZs showed leadership effects. However many counties Nanchang Prefecture KDZ 0.0110 0.0000 0.0002 0.0000 0.0028 Pengze Prefecture KDZ −0.0006 −0.0383 0.0100 0.0036 −0.0063 in KDZ showed a positive leadership effect. Qingshanhu District KDZ 0.0000 0.0000 0.0000 0.0000 0.0000 Qingyunpu District KDZ 0.0000 0.0000 0.0000 −0.0007 −0.0002 3.4. Trends in GMLPI and its decompositions Ruichang City KDZ 0.0894 0.0194 0.0129 0.0073 0.0323 Xihu District KDZ 0.0000 0.0000 0.0000 0.0000 0.0000 Fig. 2 shows the cumulative trend in the GMLPI as well as its decom- Xinjian Prefecture KDZ 0.1089 −0.0007 0.0075 −0.0046 0.0278 Xunyang District KDZ 0.0000 −0.0033 −0.0079 0.0000 −0.0028 positions for the PLEEZ as a whole. An upward trend of the GMLPI is ob- Yushui District KDZ 0.0102 −0.0021 0.0000 0.0000 0.0020 served with an average growth rate exceeding 5% during 2009–2013. Yuehu District KDZ 0.0033 0.0642 0.0044 0.0000 0.0180 Meanwhile, there is a remarkable upward trend in GTC, indicating Zhangshu City KDZ 0.0254 −0.0879 0.0249 0.0120 −0.0064 Zhushan District KDZ −0.0669 −0.1406 −0.1021 −0.0107 −0.0801 Dean Prefecture MAPZ −0.0191 −0.1524 −0.0301 −0.0938 −0.0739 Dongshan Prefecture MAPZ 0.0326 −0.0571 −0.2455 −0.1793 −0.1123 Duchang Prefecture MAPZ 0.0021 −0.1504 0.0329 −0.1312 −0.0617 Jinxian Prefecture MAPZ −0.0132 −0.1623 −0.3354 −0.1168 −0.1569 Poyang Prefecture MAPZ −0.0640 −0.2496 −0.0382 −0.1298 −0.1204 Wannian Prefecture MAPZ −0.0497 0.0144 −0.0353 −0.0377 −0.0271 Xingan Prefecture MAPZ −0.0405 −0.0713 −0.0298 −0.0341 −0.0439 Yongxiu Prefecture MAPZ −0.0504 −0.1781 −0.0274 −0.1352 −0.0978 Yugan Prefecture MAPZ 0.0081 −0.0556 −0.0528 −0.0728 −0.0433 Yujiang Prefecture MAPZ −0.0377 −0.4174 −0.0684 0.0255 −0.1245 Anyi Prefecture KEFZ 0.0400 0.0073 0.0094 −0.0397 0.0043 Fuliang Prefecture KEFZ −0.0214 −0.0262 −0.0404 −0.0181 −0.0265 Wanli District KEFZ −0.2367 −0.9453 −0.2180 −0.0656 −0.3664 Wuning Prefecture KEFZ 0.0156 −0.1623 −0.0450 0.0601 −0.0329 Xingzi Prefecture KEFZ −0.3481 −0.0793 −0.0233 0.0295 −0.1053 Mean PLEEZ −0.0122 −0.0761 −0.0294 −0.0236 −0.0353 Fig. 3. GMLPI trends for the three major functional zones.

Please cite this article as: Yu, Y., et al., Environmental catching-up, eco-innovation, and technological leadership in China's pilot ecological civilization zones, Technol. Forecast. Soc. Change (2016), http://dx.doi.org/10.1016/j.techfore.2016.05.010 Y. Yu et al. / Technological Forecasting & Social Change xxx (2016) xxx–xxx 7

Fig. 4. GEC trends of the three major functional zones. Fig. 6. MGC trends for the three major functional zones. that technological change was the main source of environmental pro- Combining the trends of the GEC, GTC, and MGC for three zones, the ductivity growth in the PLEEZ. We can conclude that environmental GMLPI appears to be largely connected to technological change. This productivity is primarily driven by eco-innovation. The downward suggests that the environmental productivity growth of the three trend in both GEC and MGC is investigated in Fig. 2, which indicates a zones also benefited from eco-innovation during this period. lack of eco-friendly catch-up and technological leadership effect in the PLEEZ. 3.5. Group innovators and meta-innovators in PLEEZ We now compare the GMLPI and its decompositions across different major functional zones. Fig. 3 shows the trends of the GMLPI for the There are two types of innovators: group innovators and three major types of functional zones of the PLEEZ. The KDZs show metafrontier innovators. Time t denotes period t in which a region values above zero and a continuously increasing trend in the GMLPI participates in constructing the production frontier. According to during 2009–2013. The MAPZs show an M-type trend, reflecting the Zhang and Wei (2015), such a region should satisfy the following fluctuations in the environmental productivity change. The KEFZs conditions: have a relatively flat trend near zero, indicating that the environmental productivity in this area remained almost unchanged. TCN0 ð9aÞ The components of the GMLPI for the three major functional zones should be also investigated. First, the trends in GEC for the three zones !t tþ1; tþ1; tþ1 b are shown in Fig. 4. The GEC index of all three zones during D x y b 0 ð9bÞ 2009–2011 shows downward trends below zero, indicating that the dis- tance between the observations and environmental technology frontier !tþ1 tþ1; tþ1; tþ1 : has grown. After 2011, the GEC of the three zones began to increase. The D x y b ¼ 0 ð9cÞ MAPZs and the KEFZs, with greater catch-up effect, showed efficiency gains in 2012, but the KEFZs and the KDZs exhibited falling GEC again Eq. (9a) suggests that the contemporaneous environmental technol- in 2013. ogy frontier should shift toward intertemporal environmental technol- Fig. 5 shows that GTC index of the three zones is above 0 during ogy, signaling the occurrence of eco-innovation. Eq. (9b) indicates that 2009–2013, indicating that eco-innovation occurred in the three the production activity of innovative regions in period t + 1 should be zones. The KEFZs showed significant eco-innovation before 2011, a outside the frontier of period t.Eq.(9c) implies that the innovative re- trend that became slow after 2011. The KDZs also show a similar pattern gion should be located on the environmental technology frontier in pe- in GTC. The MAPZs exhibit an M-shaped pattern in GTC, reflecting fluc- riod t + 1. tuations in eco-technological change in this zone. Two additional conditions are required to choose metafrontier inno- In Fig. 6,MGCfluctuations of the KDZs is near zero, indicating that vative regions: the MGC of the KDZs remained unchanged, suggesting that the technol- ogy gap of the KDZs is the same. In the MAPZs and the KEFZs during MGCN0 ð10aÞ 2009–2011, the technology gap grew; while after 2011, MGC showed an upward trend, indicating that the technology gap of these two zones fell, especially for the KEFZs. Table 7 Group innovators in the three zones.

Period Group innovator

KDZ MAPZ KEFZ

2009–2010 Donghu District Jinxian Prefecture Fuliang Prefecture Xihu District Poyang Prefecture Wanli District Yujiang Prefecture Xingzi Prefecture 2010–2011 Yushui District Jinxian Prefecture Fuliang Prefecture Zhushan District Poyang Prefecture Wanli District Yujiang Prefecture Wuning Prefecture 2011–2012 Gongqincheng City Jinxian Prefecture Wanli District Xihu District Poyang Prefecture Xunyang District Zhushan District 2012–2013 Zhushan District Jinxian Prefecture Wanli District Wannian Prefecture Fig. 5. GTC trends for the three major functional zones.

Please cite this article as: Yu, Y., et al., Environmental catching-up, eco-innovation, and technological leadership in China's pilot ecological civilization zones, Technol. Forecast. Soc. Change (2016), http://dx.doi.org/10.1016/j.techfore.2016.05.010 8 Y. Yu et al. / Technological Forecasting & Social Change xxx (2016) xxx–xxx

Table 8 county level data for the period 2009–2013. By considering regional Kruskal–Wallis test for three zones. heterogeneity and the non-radial Luenberger productivity indicator, Test Null hypothesis (H0) Chi-square p-Value we presented the global metafrontier Luenberger productivity indicator (GMLPI) for measuring environmental productivity growth and its de- GMLPI KDZ = MAPZ = KEFZ 4.17 0.124 GEC KDZ = MAPZ = KEFZ 2.98 0.225 compositions including eco-catching-up and eco-innovation. Results GTC KDZ = MAPZ = KEFZ 8.46 0.015 show that environmental productivity growth has increased by 8.71%, MGC East = Central = West 41.48 0.000 primarily driven by eco-technological change, implying that the estab- lishment of the PLEEZ is effective in encouraging eco-innovation in the PLEEZE. However, the PLEEZ lacks eco-leadership effects because the ! D G xtþ1; ytþ1; btþ1 ¼ 0 ð10bÞ metafrontier gap has not been reduced. The results also show that there are significant heterogeneities among the three major types of functional zones in the PLEEZ in environmental productivity growth Eq. (10a) states that a metafrontier innovator should be among the and its patterns. The KDZ areas showed the highest environmental pro- technological leaders, implying a decrease in the gap between the ductivity growth and technological leadership change. group frontier and the metafrontier. Condition (10b) suggests that a This empirical study has some limitations. Technically, the method metafrontier innovative region should be located along the global envi- shows a lack of statistical inference in the GMLPI. By using the ronmental technology frontier. bootstrapping method, statistical analysis of the productivity result Table 7 lists the innovators for every second year. No regions appear could be included, making the results more robust. to satisfy the conditions of metafrontier innovators. This result confirms Another limitation is that no statistical evidence is provided to sup- the findings in the previous sections: the lack of leadership effect in the port the policy implications because of the lack of data. In the future, if PLEEZ. For group innovators, in the KDZs, Zhushan District and Xihu Dis- the data is available, econometric models should be used to investigate trict are found to be innovators twice, respectively. In the MAPZs, the factors affecting environmental productivity in the PLEEZ. In addi- Jinxian County constructed the group frontier in each period. In the tion, since county level data of the PLEEZ is used in this study, the KEFZs, Wanli County appeared four times as the group innovator. metafrontier can only partially address the heterogeneity problem. Finally, to investigate any statistically significant differences among Given that firm-level data is available, this data should be used instead the three major zones in terms of the GMLPI and its decompositions, in the future. the non-parametric Kruskal–Wallis test is used, with the results shown in Table 8. The statistical test supports the hypothesis that, GTC Acknowledgements and MGC, group-specific components, for the three major zones differ from each other significantly. On the other hand, the result failed to We thank the financial support provided by the National Natural Sci- prove that the GMLPI and GEC, statistically have differences. It is ence Foundation of China (41461118), National Social Science Founda- found that the sources of regional heterogeneities are driven from tion of China (15ZDA054), Humanities and Social Science Fund of technological-related terms, i.e. innovation and technological leader- Jiangxi (JJ1420), and the China Postdoctoral Science Foundation ship in eco-development. The kernel density plot in Fig. 7 also implies (2015T80684). obvious differences in the distribution pattern among the three major zones. In addition, the related kernel density test verifies significant dif- Appendix A ferences in the distribution pattern. The Poyang Lake is the largest freshwater lake in China. Situated in 4. Conclusion the North of China's Jiangxi Province, the lake drains directly into the middle and lower reaches of the Yangtze River, thereby acting as a nat- This study investigated the effect of the establishment of the PLEEZ ural flood-buffer for the entire lower Yangtze hydrological system. His- on regional environmental total-factor productivity growth using torically the Poyang Lake experienced devastating degradation of its

Fig. 7. Kernel density estimation of GMLPI and decompositions for three zones.

Please cite this article as: Yu, Y., et al., Environmental catching-up, eco-innovation, and technological leadership in China's pilot ecological civilization zones, Technol. Forecast. Soc. Change (2016), http://dx.doi.org/10.1016/j.techfore.2016.05.010 Y. Yu et al. / Technological Forecasting & Social Change xxx (2016) xxx–xxx 9 natural environment. In the 1980s the Jiangxi Provincial Government rec- Wang, Q., Zhao, Z., Zhou, P., Zhou, D., 2013. Energy efficiency and production technology heterogeneity in China: a meta-frontier DEA approach. Econ. Model. 35, 283–289. ognized the threats and decided to take actions to protect and to rehabil- Wang, Q., Su, B., Sun, J., Zhou, P., Zhou, D., 2015. Measurement and decomposition of itate the natural environment of the province. In 1985 the Provincial energy-saving and emissions reduction performance in Chinese cities. Appl. Energy Government set up a Mountain–River–Lake Sustainable Development 151, 85–92. Weber, W., Domazlicky, B., 2001. Productivity growth and pollution in state manufactur- Program in order to tackle these ecological problems. The Poyang Lake ing. Rev. Econ. Stat. 83, 195–199. Eco-economic Zone is a further development plan of the Mountain– Wu, Y., 2009. China’s Capital Stock Series by Region and Sector. The University of Western River–Lake Sustainable Development Program by considering new chal- Australia Discussion Paper. lenges. On December 12th 2009 the State Council of China approved the Wu, T., Wang, Y., 2015. Did the establishment of Poyang Lake eco-economic zone increase agricultural labor productivity in Jiangxi Province, China? Sustainability 8 (8), 1–11. Poyang Lake Eco-economic Zone Plan as part of its National Strategy. Yörük, B.K., Zaim, O., 2005. Productivity growth in OECD countries: a comparison with This plan basically includes all three aspects of a sustainable development Malmquist indices. J. Comp. Econ. 33, 401–420. program: environmental protection, social development and economic Young, A., 2003. Gold into base metals: productivity growth in the People's Republic of China during the reform period. J. Polit. Econ. 111, 1220–1261. growth. The Poyang Lake Eco-economic Zone is located in the North of Zhang, N., Choi, Y., 2013a. Total-factor carbon emission performance of fossil fuel power Jiangxi Province, covering a total area of 51,200 km2 (app. 30% of the prov- plants in China: a metafrontier non-radial Malmquist index analysis. Energy Econ. ince). The area covers three large cities and 38 counties with a population 40, 549–559. N Zhang, N., Choi, Y., 2013b. A comparative study of dynamic changes in CO2 emission per- of 20 million (app. half of the province's population). formance of fossil fuel power plants in China and Korea. Energy Policy 62, 324–332. The entire Eco-economic Zone consists of three sub-zones. The core Zhang,N.,Choi,Y.,2014.A note on the evolution of directional distance function and its protection zone (5180 km2) includes the lake's water body up to its max- development in energy and environmental studies 1997–2013. Renew. Sust. Energ. Rev. 33, 50–59. imum size and the directly connected wetland. Environmental protection Zhang, N., Wei, X., 2015. Dynamic total factor carbon emissions performance changes in will be the high priority here; consequently economic activities have to the Chinese transportation industry. Appl. Energy 146, 409–420. follow rigorous regulations. The lakeshore zone surrounds the core pro- Zhang, C., Liu, H., Bressers, H., Buchanan, K., 2011. Productivity growth and environmental 2 regulations — accounting for undesirable outputs: analysis of China's thirty provincial tection zone by a radius of 3 km covering an area of 3750 .km .Strictreg- regions using the Malmquist–Luenberger index. Ecol. Econ. 70, 2369–2379. ulations are applied to limit all economic development in this sub-zone. Zhang, N., Kong, F., Choi, Y., Zhou, P., 2014. The effect of size-control policy on unified energy The development zone covers the largest part of the Poyang Lake Eco- and carbon efficiency for Chinese fossil fuel power plants. Energy Policy 70, 193–200. economic Zone, with an area of 42.200 km2. Economic development Zhou, P., Ang, B.W., Wang, H., 2012. Energy and CO2 emission performance in electricity generation: a non-radial directional distance function approach. Eur. J. Oper. Res. 221, and growth are a priority in this area; however all economic development 625–635. activities in this zone follow environment-friendly rules. The Poyang Lake Eco-economic Zone is a strategic development plan Yanni Yu is an Associate Professor at the Institute of Resource, Environment and Sustain- to improve Jiangxi Province's sustainable development. In order to able Development (IRESD) in the College of Economics, Jinan University. She received her PhD in International Trade from Inha University. Her current research includes Sustain- achieve this goal, the implementation of this plan splits up into short ability and CSR. Her research outputs have been published in major international journals term (2009–2015) and long term stages (2016–2020). such as Technological Forecasting and Social Change, Computational Economics, Ecological In- dicators, Social Science Journal, Chinese management studies and Sustainability. References Wenjie Wu is an Associate Professor at Heriot-Watt University, and formerly a Lecturer in Urban Studies, the University of Glasgow. He remains as a research affiliate at the Spatial Boussemart, J.P., Briec, W., Kerstens, K., Poutineau, J.C., 2003. Luenberger and Malmquist Economic Research Centre, London School of Economics and Political Science, and a co- productivity indices: theoretical comparisons and empirical illustration. Bull. Econ. investigator of the ESRC-funded Urban Big Data Center (Glasgow). His research focuses Res. 55, 391–405. on geographical implications of urban transformations in China. His work has been pub- Chen, S., Golley, J., 2014. ‘Green’ productivity growth in China's industrial economy. Ener- lished in journals such as Annals of the Association of American Geographers, Journal of Re- gy Econ. 44, 89–98. gional Science, Papers in Regional Science,andamongothers. Choi, Y., Oh, D., Zhang, N., 2015. Environmentally sensitive productivity growth and its de- compositions in China: a metafrontier Malmquist–Luenberger productivity index ap- Tao Zhang is a Lecturer in Marketing and Sustainability in Birmingham Business School, proach. Empir. Econ. 49, 1017–1043. the University of Birmingham. He gained his PhD in Energy Economics from the Energy Chung, Y.H., Färe, R., Grosskopf, S., 1997. Productivity and undesirable outputs: a direc- Policy Research Group, Judge Business School, University of Cambridge. His research inter- tional distance function approach. J. Environ. Manag. 51, 229–240. ests are in the areas of energy economics and policy, energy consumer behaviour and in- Färe, R., Grosskopf, S., 2005. New directions: efficiency and productivity. Springer, New novation management, and agent-based modeling for the energy market. His publications York. appear in Technological Forecasting and Social Change, Energy and Buildings, Energy Policy, Färe, R., Grosskopf, S., Lovell, C.A.K., Pasurka, C., 1989. Multilateral productivity compari- and the Journal of Product Innovation Management,etc. sons when some outputs are undesirable: a nonparametric approach. Rev. Econ. Stat. 71, 90–98. Yanchu Liu is an Assistant Professor of Finance at Lingnan College, Sun. Yat-sen University. Färe, R., Grosskopf, S., Norris, M., Zhang, Z., 1994. Productivity growth, technical progress, He received a Ph.D. degree in Financial Engineering at the Chinese University of Hong Kong and efficiency change in industrialized countries. Am. Econ. Rev. 84, 66–83. in 2012 and got his M.S. and B.S. degrees in Statistics at the University of Science and Tech- Färe, R., Grosskopf, S., Pasurka Jr., C.A., 2001. Accounting for air pollution emissions in nology of China in 2008 and 2005, respectively. His main research interests are in financial measures of state manufacturing productivity growth. J. Reg. Sci. 41, 381–409. economics and related computational problems with applications. His work has been pub- Fujii, H., Managi, S., Matousek, R., 2014. Indian bank efficiency and productivity changes lished in journals such as Operations Research and Applied Energy. with undesirable outputs: a disaggregation approach. J. Bank. Financ. 38, 41–50. Kumar, S., 2006. Environmentally sensitive productivity growth: a global analysis using Malmquist–Luenberger index. Ecol. Econ. 56, 280–293.

Li, K., Lin, B., 2015. Metafroniter energy efficiency with CO2 emissions and its convergence analysis for China. Energy Econ. 48, pp. 230–241.

Please cite this article as: Yu, Y., et al., Environmental catching-up, eco-innovation, and technological leadership in China's pilot ecological civilization zones, Technol. Forecast. Soc. Change (2016), http://dx.doi.org/10.1016/j.techfore.2016.05.010